[HADOOP-18562]: S3A: support custom S3 and STS headers.

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/Constants.java

@@ -775,6 +775,29 @@ private Constants() {
       "fs.s3a." + Constants.AWS_SERVICE_IDENTIFIER_STS.toLowerCase()
           + ".signing-algorithm";
 
+  /** Prefix for S3A client-specific properties. */
+  public static final String FS_S3A_CLIENT_PREFIX = "fs.s3a.client.";
+
+  /** Custom headers postfix */
+  public static final String CUSTOM_HEADERS_POSTFIX = ".custom.headers";
+
+  /**
+   * List of custom headers to be set on the service client.
+   * Multiple parameters can be used to specify custom headers.
+   * fs.s3a.client.s3.custom.headers - headers to add on all the s3 requests.
+   * fs.s3a.client.sts.custom.headers - headers to add on all the sts requests.
+   * Examples
+   * CustomHeader {@literal ->} 'Header1:Value1'
+   * CustomHeaders {@literal ->} 'Header1=Value1:Value2,Header2=Value1'
+   */
+  public static final String CUSTOM_HEADERS_STS =
+      FS_S3A_CLIENT_PREFIX + Constants.AWS_SERVICE_IDENTIFIER_STS.toLowerCase()
+          + CUSTOM_HEADERS_POSTFIX;
+
+  public static final String CUSTOM_HEADERS_S3 =
+      FS_S3A_CLIENT_PREFIX + Constants.AWS_SERVICE_IDENTIFIER_S3.toLowerCase()
+          + CUSTOM_HEADERS_POSTFIX;
+
   public static final String S3N_FOLDER_SUFFIX = "_$folder$";
   public static final String FS_S3A_BLOCK_SIZE = "fs.s3a.block.size";
   public static final String FS_S3A = "s3a";

hadoop-tools/hadoop-aws/src/main/java/org/apache/hadoop/fs/s3a/impl/AWSClientConfig.java

@@ -22,6 +22,9 @@
 import java.net.URI;
 import java.net.URISyntaxException;
 import java.time.Duration;
+import java.util.Arrays;
+import java.util.List;
+import java.util.Map;
 import java.util.concurrent.TimeUnit;
 
 import org.slf4j.Logger;
@@ -72,6 +75,8 @@
 import static org.apache.hadoop.fs.s3a.Constants.SIGNING_ALGORITHM_STS;
 import static org.apache.hadoop.fs.s3a.Constants.SOCKET_TIMEOUT;
 import static org.apache.hadoop.fs.s3a.Constants.USER_AGENT_PREFIX;
+import static org.apache.hadoop.fs.s3a.Constants.CUSTOM_HEADERS_S3;
+import static org.apache.hadoop.fs.s3a.Constants.CUSTOM_HEADERS_STS;
 import static org.apache.hadoop.fs.s3a.impl.ConfigurationHelper.enforceMinimumDuration;
 import static org.apache.hadoop.fs.s3a.impl.ConfigurationHelper.getDuration;
 import static org.apache.hadoop.util.Preconditions.checkArgument;
@@ -116,6 +121,8 @@ public static ClientOverrideConfiguration.Builder createClientConfigBuilder(Conf
 
     initUserAgent(conf, overrideConfigBuilder);
 
+    initRequestHeaders(conf, overrideConfigBuilder, awsServiceIdentifier);
+
     String signer = conf.getTrimmed(SIGNING_ALGORITHM, "");
     if (!signer.isEmpty()) {
       LOG.debug("Signer override = {}", signer);
@@ -407,6 +414,36 @@ private static void initSigner(Configuration conf,
     }
   }
 
+  /**
+   *
+   * @param conf hadoop configuration
+   * @param clientConfig client configuration to update
+   * @param awsServiceIdentifier service name
+   */
+  private static void initRequestHeaders(Configuration conf,
+      ClientOverrideConfiguration.Builder clientConfig, String awsServiceIdentifier) {
+    String configKey = null;
+    switch (awsServiceIdentifier) {
+      case AWS_SERVICE_IDENTIFIER_S3:
+        configKey = CUSTOM_HEADERS_S3;
+        break;
+      case AWS_SERVICE_IDENTIFIER_STS:
+        configKey = CUSTOM_HEADERS_STS;
+        break;
+      default:
+        // Nothing to do. The original signer override is already setup
+    }
+    if (configKey != null) {
+      Map<String, String> awsClientCustomHeadersMap =
+              S3AUtils.getTrimmedStringCollectionSplitByEquals(conf, configKey);
+      awsClientCustomHeadersMap.forEach((header, valueString) -> {
+        List<String> headerValues = Arrays.asList(valueString.split(":"));
+        clientConfig.putHeader(header, headerValues);
+      });
+      LOG.debug("headers for {} client = {}", awsServiceIdentifier, clientConfig.headers());
+    }
+  }
+
   /**
    * Configures request timeout in the client configuration.
    * This is independent of the timeouts set in the sync and async HTTP clients;

hadoop-tools/hadoop-aws/src/test/java/org/apache/hadoop/fs/s3a/impl/TestAwsClientConfig.java

@@ -18,6 +18,7 @@
 
 package org.apache.hadoop.fs.s3a.impl;
 
+import java.io.IOException;
 import java.time.Duration;
 import java.util.Arrays;
 
@@ -29,11 +30,16 @@
 
 import org.apache.hadoop.conf.Configuration;
 import org.apache.hadoop.test.AbstractHadoopTestBase;
+import org.apache.hadoop.util.Lists;
 
+import static org.apache.hadoop.fs.s3a.Constants.AWS_SERVICE_IDENTIFIER_S3;
+import static org.apache.hadoop.fs.s3a.Constants.AWS_SERVICE_IDENTIFIER_STS;
 import static org.apache.hadoop.fs.s3a.Constants.CONNECTION_ACQUISITION_TIMEOUT;
 import static org.apache.hadoop.fs.s3a.Constants.CONNECTION_IDLE_TIME;
 import static org.apache.hadoop.fs.s3a.Constants.CONNECTION_KEEPALIVE;
 import static org.apache.hadoop.fs.s3a.Constants.CONNECTION_TTL;
+import static org.apache.hadoop.fs.s3a.Constants.CUSTOM_HEADERS_S3;
+import static org.apache.hadoop.fs.s3a.Constants.CUSTOM_HEADERS_STS;
 import static org.apache.hadoop.fs.s3a.Constants.DEFAULT_CONNECTION_ACQUISITION_TIMEOUT_DURATION;
 import static org.apache.hadoop.fs.s3a.Constants.DEFAULT_CONNECTION_IDLE_TIME_DURATION;
 import static org.apache.hadoop.fs.s3a.Constants.DEFAULT_CONNECTION_KEEPALIVE;
@@ -48,6 +54,7 @@
 import static org.apache.hadoop.fs.s3a.Constants.REQUEST_TIMEOUT;
 import static org.apache.hadoop.fs.s3a.Constants.SOCKET_TIMEOUT;
 import static org.apache.hadoop.fs.s3a.impl.AWSClientConfig.createApiConnectionSettings;
+import static org.apache.hadoop.fs.s3a.impl.AWSClientConfig.createClientConfigBuilder;
 import static org.apache.hadoop.fs.s3a.impl.AWSClientConfig.createConnectionSettings;
 import static org.apache.hadoop.fs.s3a.impl.ConfigurationHelper.enforceMinimumDuration;
 
@@ -201,4 +208,56 @@ public void testCreateApiConnectionSettingsDefault() {
   private void setOptionsToValue(String value, Configuration conf, String... keys) {
     Arrays.stream(keys).forEach(key -> conf.set(key, value));
   }
+
+  /**
+   * if {@link org.apache.hadoop.fs.s3a.Constants#CUSTOM_HEADERS_STS} is set,
+   * verify that returned client configuration has desired headers set.
+   */
+  @Test
+  public void testInitRequestHeadersForSTS() throws IOException {
+    final Configuration conf = new Configuration();
+    conf.set(CUSTOM_HEADERS_STS, "foo=bar:baz,qux=quux");
+    Assertions.assertThat(conf.get(CUSTOM_HEADERS_S3))
+            .describedAs("Custom client headers for s3 %s", CUSTOM_HEADERS_S3)
+            .isNull();
+
+    Assertions.assertThat(createClientConfigBuilder(conf, AWS_SERVICE_IDENTIFIER_S3).headers().size())
+            .describedAs("Count of S3 client headers")
+            .isEqualTo(0);
+    Assertions.assertThat(createClientConfigBuilder(conf, AWS_SERVICE_IDENTIFIER_STS).headers().size())
+            .describedAs("Count of STS client headers")
+            .isEqualTo(2);
+    Assertions.assertThat(createClientConfigBuilder(conf, AWS_SERVICE_IDENTIFIER_STS).headers().get("foo"))
+            .describedAs("STS client 'foo' header value")
+            .isEqualTo(Lists.newArrayList("bar", "baz"));
+    Assertions.assertThat(createClientConfigBuilder(conf, AWS_SERVICE_IDENTIFIER_STS).headers().get("qux"))
+            .describedAs("STS client 'qux' header value")
+            .isEqualTo(Lists.newArrayList("quux"));
+  }
+
+  /**
+   * if {@link org.apache.hadoop.fs.s3a.Constants#CUSTOM_HEADERS_S3} is set,
+   * verify that returned client configuration has desired headers set.
+   */
+  @Test
+  public void testInitRequestHeadersForS3() throws IOException {
+    final Configuration conf = new Configuration();
+    conf.set(CUSTOM_HEADERS_S3, "foo=bar:baz,qux=quux");
+    Assertions.assertThat(conf.get(CUSTOM_HEADERS_STS))
+            .describedAs("Custom client headers for STS %s", CUSTOM_HEADERS_STS)
+            .isNull();
+
+    Assertions.assertThat(createClientConfigBuilder(conf, AWS_SERVICE_IDENTIFIER_STS).headers().size())
+            .describedAs("Count of STS client headers")
+            .isEqualTo(0);
+    Assertions.assertThat(createClientConfigBuilder(conf, AWS_SERVICE_IDENTIFIER_S3).headers().size())
+            .describedAs("Count of S3 client headers")
+            .isEqualTo(2);
+    Assertions.assertThat(createClientConfigBuilder(conf, AWS_SERVICE_IDENTIFIER_S3).headers().get("foo"))
+            .describedAs("S3 client 'foo' header value")
+            .isEqualTo(Lists.newArrayList("bar", "baz"));
+    Assertions.assertThat(createClientConfigBuilder(conf, AWS_SERVICE_IDENTIFIER_S3).headers().get("qux"))
+            .describedAs("S3 client 'qux' header value")
+            .isEqualTo(Lists.newArrayList("quux"));
+  }
 }